MAGAR package overview

We developed “Methylation-Aware Genotype Association in R” (MAGAR) as a new computational framework to determine methQTLs from DNA methylation and genotyping data. MAGAR supports both sequencing-based assays including whole-genome (bisulfite) sequencing and microarray-based data. It is the first computational framework for performing methQTL analysis starting from raw DNA methylation and genotyping microarray data. The pipeline implemented within MAGAR comprises the following phases:

Data import and preprocessing using established software packages such as PLINK [32], RnBeads [30, 31], and CRLMM [35, 36]. Additional modules for quality control and standard processing using these packages are available to the user. MAGAR supports automated imputation using the Michigan Imputation Server [54].

MethQTL calling, i.e., computing associations between genotype and a DNA methylation state. A two-stage approach is employed: (i) Define CpG correlation blocks as groups of CpGs that are highly correlated across the samples to mimic DNA methylation haplotypes. (ii) From each of these correlation blocks, a tag-CpG is selected as a representative of the block and associations are computed with all SNPs up to a given distance using either a linear modeling strategy or using external software tools (e.g., fastQTL [28]). All SNP-CpG pairs that have a P-value below a user-defined cutoff are returned.